Summarize REMAP item-level residuals for an ROI

Extract per-item distances between memory and perception prototypes in the jointly-whitened space, before (naive) and after the REMAP correction. Works from a `regional_mvpa_result` when `return_fits=TRUE`, or directly from a predictor list returned by `process_roi()`.

Usage

summarize_remap_items(x, roi = NULL)

Arguments

x

Either a `regional_mvpa_result` (preferred) or a predictor list containing `diag_by_fold`.

roi

ROI index or id when `x` is a regional result. If `NULL` and `x` is a predictor, it is ignored.

Value

A tibble with columns: `item`, `res_naive`, `res_remap`, `res_ratio`, and `n_folds`.

Examples

# A predictor list as stored in regional_mvpa_result$fits when
# run_regional(..., return_fits = TRUE) is used with remap_rrr_model():
pred <- list(diag_by_fold = list(
  list(train_items = c("a", "b"), item_res_naive = c(4, 8),
       item_res_remap = c(2, 4)),
  list(train_items = c("a", "b"), item_res_naive = c(6, 10),
       item_res_remap = c(3, 5))
))
summarize_remap_items(pred)
#> # A tibble: 2 × 5
#>   item  res_naive res_remap n_folds res_ratio
#>   <chr>     <dbl>     <dbl>   <int>     <dbl>
#> 1 a             5       2.5       2       0.5
#> 2 b             9       4.5       2       0.5